Table 1.
Software for feature prediction.
Figure 1.
Make up of FFPred 2.0 GO term list and differences from the previous release.
High-level summary of the FFPred 2.0 GO term vocabulary: each tag in the cloud is a child of the root of the MF or BP domains (shown in orange and blue respectively), and their size reflects how many of their descendants can be predicted (A). Extent of the overlap between FFPred 1.0 and FFPred 2.0 GO term lists in the MF (B) and BP (C) domains.
Table 2.
Comparison of MCC values for common GO terms.
Figure 2.
Comparison of FFPred 1.0 and FFPred 2.0 by means of precision-recall plots.
Precision-recall plots comparing the performance of FFPred 1.0 and FFPred 2.0 on a common dataset of 148 human proteins, for predictions in the MF domain (left) and BP domain (right).
Figure 3.
Comparison of FFPred 2.0 and BLAST-based tools by means of precision-recall plots.
Precision-recall plots comparing the performance of FFPred 2.0 to the BLAST-based tools described in the main text on a common dataset of 148 human proteins, for predictions in the MF domain (left) and BP domain (right).
Figure 4.
Evaluation of FFPred 2.0 on eukaryotic proteomes by means of precision-recall plots.
Precision-recall plots describing the performance of FFPred 2.0 on several eukaryotic proteomes, for predictions in the MF domain (left) and BP domain (right).
Figure 5.
Sample FFPred 2.0 output within the PSIPRED web server.
Top part of the FFPred 2.0 tab within the results page for the submission of the sequence of Serotonin receptor 1B to the PSIPRED Protein Analysis Workbench web server. Predicted GO term assignments are listed together with the associated posterior probability and SVM reliability level.